摘要： 提出一种基于逼近信号统计特征和线性伸缩恢复技术的音频数字零水印算法。对原始水印采用重复码进行纠错编码，将密钥通过伪随机序列发生器产生与重复编码等长的PN序列P，用P对编码后的水印进行调制以提高水印检测的正确率。检测时采用线性伸缩恢复技术来消除时间轴上线性伸缩带来的影响，并实现了盲检测。实验结果证明，该算法对于各种攻击尤其对剪切和MP3攻击具有较好的鲁棒性。

关键词: 音频数字水印, 零水印, 纠错码, 统计特征, 线性伸缩恢复

Abstract: This paper presents an audio zero-digital watermark algorithm based on statistical feature of approximation signal and the technology of restoration of the linear scaling. The original watermark is error corrected coded with repetition code. In order to enhance the correct probability of detecting, the recoded watermark is modulated with PN sequence P which is produced by the key through Pseudo-random sequences generator, with the same length of the recoded code. During the process of detecting, the technology of restoration of the linear scaling is used to eliminate the influence brought by linear scaling on time axis, and a blind detecting is realized. Experiment confirms the expected high robustness to many kinds of attacks especially to cutting-attack and MP3-attack.

Key words: audio digital watermark, zero watermark, error correction code, statistical feature, restoration of the linear scaling

中图分类号: 

• TP309.2